Clemson UniversityTigerPrints

All Theses Theses

5-2018

Tenure Security and Housing Investment: A Studyof Bangladeshi Poor SettlementsThu Anh NguyenClemson University, anhthu746583@gmail.com

Follow this and additional works at: https://tigerprints.clemson.edu/all_theses

This Thesis is brought to you for free and open access by the Theses at TigerPrints. It has been accepted for inclusion in All Theses by an authorizedadministrator of TigerPrints. For more information, please contact kokeefe@clemson.edu.

Recommended CitationNguyen, Thu Anh, "Tenure Security and Housing Investment: A Study of Bangladeshi Poor Settlements" (2018). All Theses. 2867.https://tigerprints.clemson.edu/all_theses/2867

TENURE SECURITY AND HOUSING INVESTMENT:

A STUDY OF BANGLADESHI POOR SETTLEMENTS

A Thesis

Presented to

the Graduate School of

Clemson University

by

Thu Anh Nguyen

May 2018

Accepted by

Dr. James H. Spencer, Committee

Chair Dr. Timothy Green

Dr. Elora Raymond

In Partial Fulfillmentof the Requirements for the Degree

Master of City and Regional Planning

ii

ABSTRACT

The purpose of this document is to explore the relationship between land tenure security

and housing investment among households in Bangladeshi poor settlements. Tenure

security is viewed through ownership over the property. Whether people squat, rent, or

own their living space indicates a certain level of tenure security. Level of housing

investment is approached through the physical structure of the house, or more specifically,

floor materials. Cluster-robust standard error logistic regression method is utilized to

explore this relationship in order to fix the issue of households located in the same

settlement being correlated with each other. Households with greater tenure security, either

renters or owner occupants, are more likely than squatters to have cement floors, which are

more permanent and expensive materials then others like bamboo, wood or sand.

However, renters of non-government properties are more likely than owner occupants to

use cement floors. After all, although greater tenure security does have a positive impact

on housing investment, the market force in which housing is seen as a business opportunity

can be a stronger factor in encouraging investment even in poor communities.

iii

ACKNOWLEDGMENTS

I would first like to thank my awesome advisor, Dr. James Spencer. Without him, I

wouldn’t have come to Clemson and had my great time here. His consistent encouragement

and support steered me in the right direction and complete this thesis.

I would also like to thank my committee members: Dr. Timothy Green, and Dr. Elora

Raymond for giving me valuable feedbacks and supports to finish this year long project.

I would also like to acknowledge the Bangladesh National Urban Poverty Reduction

Program team: John, Palash, and many others for allowing me to have one of the most

memorable summer in Bangladesh where I learned so much, started the ideas for my thesis,

and gathered the data needed. Without the NUPRP, this thesis about Bangladeshi poor

settlements wouldn’t have happened.

I would also want to send my love and thank to Laura and Lee, my Clemson mom and dad

for making Clemson a home for me from day one. Al the stories, thoughts and advice from

you not only help my stay in Clemson but will also go with me in the future.

Finally, I must express my very profound gratitude to my parents, my siblings (Bin and

Thao) and to my husband-to-be for providing me with unfailing support and continuous

encouragement throughout my years of study and through the process of researching and

writing this thesis. This accomplishment would not have been possible without them.

TABLE OF CONTENTS

Chapter 1. INTRODUCTION ............................................................................................. 1

Chapter 2. BACKGROUND ............................................................................................... 3

Chapter 3. LITERATURE REVIEW .................................................................................. 5

3.1. The values of land - Use value vs. Exchange value ............................................ 5

3.2. Land tenure - Land tenure security ..................................................................... 6

3.3. Connections between land tenure security and housing investment ................. 10

3.4. Summary of literature review ........................................................................... 15

Chapter 4. RESEARCH DESIGN .................................................................................... 16

4.1. Data Collection ................................................................................................. 16

4.2. Data Limitation ................................................................................................. 19

4.3. Conceptualization ............................................................................................. 19

4.4. Variable Descriptions........................................................................................ 22

4.5. Statistical Procedure.......................................................................................... 27

Chapter 5. RESULTS........................................................................................................ 31

5.1. Descriptive statistics ......................................................................................... 31

5.2. Cluster-robust standard errors logistic regression model.................................. 36

Chapter 6. DISCUSSION ................................................................................................. 40

Chapter 7. IMPLICATION ............................................................................................... 42

REFERENCES ................................................................................................................. 44

APPENDIX ....................................................................................................................... 47

iv

Abstract................................................................................................................................ii

Acknowledgments...............................................................................................................iii

Title Page..............................................................................................................................i

List of Figures.......................................................................................................................v

List of Tables.......................................................................................................................vi

LIST OF FIGURES

Figure 2.1. Narayanganj and Chandpur in Bangladesh ...................................................... 4

Figure 4.1. Steps of Urban Poor Settlement Mapping (NUPRP 2017) ............................ 17

Figure 4.2. Tin Houses in Bangladeshi Poor settlements ................................................. 21

Figure 4.3. Conceptual model of housing investment ...................................................... 22

v

LIST OF TABLES

Table 4.1. List of variables ............................................................................................... 25

Table 5.1. Descriptive statistics ........................................................................................ 31

Table 5.2. Floor material and Ownership type ................................................................. 32

Table 5.3. Road condition of in poor households ............................................................. 33

Table 5.4. Road condition and housing material .............................................................. 33

Table 5.5. Proximity to river and housing material .......................................................... 34

Table 5.6. Cement and proximity to the railroad .............................................................. 34

Table 5.7. Cement and the use of TV ............................................................................... 35

Table 5.8. Logistic regression with cluster-robust standards error results ....................... 38

Table 5.9. Mixed-effects logistic regression result ........................................................... 39

vi

1

Chapter 1. INTRODUCTION

Due to rapid urbanization and rural-to-urban migration trend, urban poverty has

been growing into an urgent issue of many cities, especially in the developing world. In

addition to that, poor communities are often concentrated in areas with high risks of natural

disaster, social problems or eviction (Coulton & Pandley, 1992). These risks impact

people’s ability to manage and utilize their available assets. Natural disasters cause

problems like flooding or serious waterlogging leading to health issues and burden on

everyday activities. Kids without proper education due to the lack of access to facilities

become stuck in a poverty cycle. Fear of eviction can discourage people from investing in

their housing, and improving their living conditions; and displacement itself is a major

stressor and can lead to loss of material belongings and access to jobs, social connections,

and community resources.

It is clear that environmental characteristics significantly influence people’s ability

to manage and enhance different forms of assets like labor, financial or social capital.

Moreover, effective solutions to help people break out of poverty requires good

understanding this relationship. While most existing anti-poverty policies either focus on

fixing the locational conditions or supporting individuals in terms of assets, better

interventions should result from the people-place interactions (Spencer, 2004).

I am interested in looking into these relationships between place and people, or

more specifically, place characteristics and people’s process and capacity of utilizing their

assets. In order to do that, this thesis investigates the relationship between land tenure

security and housing investment in two Bangladeshi cities: Chandpur, and Narayanganj.

2

Land tenure security is comprised of the characteristics of the property, the type of

ownership over it, and the actual perceptions of people about security. It is an important

locational aspect influencing both the use value and the exchange value of the property.

Looking at the level of investment in housing allows us to see how poor people manage

and utilize this asset of land plus housing given the context of the place they live.

This research will provide an understanding of the relationship between locational

aspects and the poor’s asset vulnerability through the lens of land tenure and housing values

supporting the search for comprehensive solutions to the issues of poverty. It should be

emphasized that effective anti-poverty action is not only about giving people the assets to

change the characteristics and values of the place they live but also about understanding

and utilizing the interactions between them.

The thesis will start with a background of the study areas. The second section is a

literature review of studies have been done from various disciplines looking at place values,

land tenure, and housing. The third section is a research design with information about data

collection and statistical method will be used. The results from this analysis will follow in

the fourth section. A discussion and an implementation section will conclude this research.

3

Chapter 2. BACKGROUND

Bangladesh is a developing South Asian country bordering India, Myanmar, Nepal,

Bhutan, and China. Despite the small area of about 130,000 sq. km (land), the rapidly

growing population of over 150 million (in 2017) makes it one of the most populous and

dense country in the world (The World Factbook, 2018). Bangladesh experiences a rapid

rate of urbanization due to natural population increase, expansion of urban areas, as well

as rural-to-urban migration. Besides being an indication of growth, urbanization has led to

various urban issues such as lack of urban services, traffic congestions, crimes, and a very

noticeable increase in slums or poor settlements with substandard living conditions. A

significant portion of the urban population live below the poverty line. In 2005, Dhaka, the

capital city, had over 35% of its population considered poor or extremely poor (Ahmed,

2007). Low-quality housing is one of the major problems that the poor have to face. Ninety-

nine percent live in poor-quality structures, and over 95% of the houses are less than 150

square feet (Ahmed, 2007).

Land tenure security has been drawing lots of attention as the effective action to

improve life of the urban poor. It is interesting to see if there is really connection between

land tenure security and housing quality in Bangladeshi poor settlements. And if yes, what

aspect of land tenure security most impacts the housing condition and how can policy

makers utilize that for future planning and development? Examples from Bangladesh can

be informative for other countries and cities in the developing world facing the same issues

4

of urbanization and poverty and looking for effective solutions regarding land tenure and

housing conditions.

Figure 2.1. Narayanganj and Chandpur in Bangladesh

The Bangladeshi cities of Chandpur and Narayanganj are shown in Figure 1 below.

These have been selected as two of the first pilot cities for the National Urban Poverty

Reduction Program by the United Nations, which collects and processes data about urban

poor settlements in Bangladesh in order to support the efforts of reducing urban poverty

and improving people’s livelihood. This would be a great source of information for me to

explore the relationship between land tenure security and housing investment, which might

5

be valuable, not only in the context of Bangladesh but also in understanding the

phenomenon in other developing countries.

Chapter 3. LITERATURE REVIEW

There have been numerous studies about land tenure security as well as housing

investment in different contexts. This literature review section looks at those previous

studies in order to explore the approaches, understanding and evidences for the interactions

between these two widely discussed topics. It will start with how land has different values

to people benefiting from using, renting, or selling. As an important aspect of a piece of

land, land tenure is an important factor to be considered. The second and third parts of this

section will explore how people understand tenure security and how that is connected with

housing investment.

3.1. The values of land - Use value vs. Exchange value

According to Walker (1981), land is one type of commodity which can be used or

exchanged in many ways leading to the two categories of use and exchange values coming

associated with each piece of land. Use value refers to the utilities that can be enjoyed by

the consumption of that commodity or use of land in this case. Use value of the same

commodity is seen differently by people. The same piece of land has different meanings

and importance to its owners. The price that someone willing to pay for the land also

reflects the use value that they think they can obtain from the land.

Exchange value is the value of land when it gets exchanged on the open market

(Clark, 1982). The price is determined by the return on investment that people who rent or

buy the land such as homeowner, landlord, realtors etc. expect to get from that decision

6

(Harvey, 2010). People who see the most benefits out of the land would be willing to pay

the highest price for the land thus manage to get it. Under this assumption, land allocation

to maximize the exchange value will also lead to the maximization of use value. However,

as Pivo (1984) noted, this can only happen in a perfectly competitive market. In situation

of monopoly or absolute rent, it is the ones who can absorb the costs and afford high rent

who will eventually have the land. Moreover, land uses also have spillover effects which

are often neglected in the consideration of use values. A beautiful garden is not only

enjoyed by its owner but can also be enjoyed by the neighbors leading to higher use values

than expected.

3.2. Land tenure - Land tenure security

From the above overview of ideas about values of land, we have seen that many

different aspects contribute to the land value. One important factor or set of factors is the

characteristics of the land itself ranging from the location, size, fertility, etc. Land tenure

is an important factor in how residents value the land.

Especially in the context of poor communities, having access to secure shelter very

important for them to enhance their livelihood. Hence the importance of land ownership

and land tenure has been widely recognized. DFID (2002) has set ensuring land tenure

security as one of the major efforts aiming at poverty reduction. But what does it mean to

have land tenure? And does land tenure actually constitute people’s security?

Emergence of property rights to land

Deininger and Feder (2001) explained the process of individualization of land

property rights as response to increase in population, technical change, and the capital

7

market growth. During the early stage of development, land was normally under communal

property system and the use of land is temporal mostly for cultivation. With the increase

in population, land become scarcer, intra-community land rental came into existence

making the tenure system more complex and requiring better definition of property rights

to land. Moreover, with the increase in technical change, more profits can be gained from

the use of land leading to higher value in land especially of certain quality asking for good

definition of rights over land. With the growth of markets for output, capital, and insurance,

having the rights to land can lead to the reduction of risk as it brings promise of reliable

income from crops or different types of off-farm income such as financial credits. Through

its long emergence and evolution, land rights can now be put into and viewed by the system

of land tenure which is now reviewed in the following section.

Land tenure – A system for property rights

Land tenure is the specified rights to the land (Payne, 2004). FAO (2002) has

defined land tenure as a system which influences the relationships among people with

respect to land specifying the rights to use, or exchange the land and responsibilities over

it. Depending on different tenure system in various areas, land tenure is classified in many

different ways. However, the classification is normally based on statutory categories like

‘freeholder’, ‘leaseholder’, and non-statutory ones like ‘squatter owner’, ‘squatter tenants’

and ‘others’ (Payne, 2004). Along with each of the statutory categories are specific rights

to the property. As a result, having tenure status is normally associated with having proper

rights to the properties thus the security over their ownership and control over the property.

8

Dales (1968) viewed land as an asset for which the different types of ownership

consist of “a bundle of legally-defined user rights.” He also presented the four major types

of ownership arrangement ranging from common-property, restricted common property,

status-tenure/fixed-tenure, and full-tenure. Common-property is the asset which can be

used by everyone. Restricted common-property put some limits on the uses of the assets.

The ownership that gives exclusive rights to use but not transfer the asset is status-

tenure/fixed-tenure. When the owner is able to use and transfer the asset, it is full

ownership.

Land types – Property rights as an institution

Feder and Feeny (1991) attempted at viewing land rights through the context of the

social and economic institutional structure, including constitutional order – rules about how

society is organized; institutional arrangement - rules specified by the constitutional order;

and normative behavioral codes - cultural values influencing behaviors. Based on the

institutions associated with the land, it is categorized into four categories which also come

along with some set of property rights. According to Feder and Feeny (1991), this

categorization is an ideal analytical tool in understanding the rights to land. Different

several different types of land can be observed in a society. Moreover, although not very

common, a single tract of land can be under more than one types. This way of looking into

land rights offer an additional approach of understanding to the land tenure system

approach.

The concepts of land tenure security

9

Effects of land titling programs have been widely investigated as a tool for

development especially for poor communities (Galiani & Schargrodsky, 2006). However,

with the complexity of the tenure system especially in developing countries, property rights

specified by the status and tenure security don’t always match as they should.

So what does it mean to have tenure security? Place et al. (1994) defined that land

tenure security is achieved when people are aware of their rights to use, benefit from the

labor spent on the land, and transfer it. This definition implies both the objective and

subjective components of land tenure security. The objective aspect of it comes from the

recognition and enforcement of rights through certain official or unofficial guarantee

through laws for example. The subjective aspect of it comes from the perceptions of the

owners about their own security.

Van Gelder (2010) proposed a tripartite approach to tenure security consisting of

de jure, de facto, and perceived tenure security. De jure or legal tenure is the most widely

mentioned aspect of tenure security. Legal property rights specified by legal status is

widely believed as an important factor for security (De Soto 2000.) With this approach,

tenure is normally thought of as legal status implying security or lack of security in terms

of tenure. Varley (2013) believed this dichotomous approach to legal tenure security is not

sufficient, especially in the context of complicated tenure systems in developing countries

where tenure. Another important aspect is de facto security, which has also been widely

studied (Nakamura, 2014). De facto security is gained by the “(f)actual situation on the

ground” rather than legal status influencing how secure people feel about control over the

property (Van Gelder & Luciano, 2015). There are intrinsic elements contributing to de

10

facto tenure security such as the length of occupancy, the size of the settlement, or the level

of community organization. Extrinsic aspects such as third-party support, or political

acceptance can also be counted into de facto security factors (Van Gelder, 2015). Finally,

one’s perceptions of security is also an important aspect of tenure security. This is the

estimate of the household about the chance of eviction they face. This perception of

eviction risk can be affected by threat from individuals or groups, conflicts with neighbors,

gangs, or family members. Moreover, without legal rights to the place they live on, they

might also feel insecure and susceptible to eviction. Thus, both de jure and de facto

contribute to perceived tenure security. However, the magnitude of each’s influence on

perceived security is not clear and dependent on a particular case. Perceptions of people

can also be affected by other factors besides de jure and de facto security. Although legal

tenure security is easier to be aware of and assessed, de facto and perceived tenure security

can be more important especially for poor communities.

3.3. Connections between land tenure security and housing investment

There have been many studies looking into the relationship between land tenure

security and housing investment and the mechanism behind the possible relationship. From

the previous studies, there are mixed results about whether improved land tenure security

lead to greater housing investment. Studying groups of poor households in outskirts Buenos

Aires, Argentina, Galiani and Schargrodsky (2010) observed a positive relationship

between having an official land title and investment. Brasselle et al. (2002), found no

effects in Burkina Faso. Seeing that there might be the problem of endogeneity as people

invest in order to get better tenure security, Besley (1995) used instrumental variable to

11

investigate this relationship in two Ghanaian regions of Wassa and Anloga. For Wassa,

improved land rights results in greater investment while there is no significant impact in

Anloga. With this mixed result, he suggests looking more into the determinants of rights

based on different types of legal, de facto, and perceptual security as well as the

mechanisms behind the relationship to understand the variation in the relationship between

land tenure and investment in housing

Risk of investment

The first mechanism that can lead to possible impacts of land tenure security on

housing is risk level of investment. Without fear of seisure or eviction, they are more

willing to spend on building up. If the property is under threat of eviction, people run the

risk of losing their investment in housing or infrastructure (Demsetz, 1974). Field (2005)

examined the impacts of the titling program in Peru. He found that the property rights come

with the title encouraged people to invest more on housing renovation. The rate of

residential renovation increased by more than two-third. In addition to that, the increase is

mainly due to the lower threat of eviction. Not only security given by official title but also

the de facto and perceived security are important in influencing the investment decision of

households. Nakamura (2016) analyzed the data about slums in Pune, India and found out

that both legal and de facto tenure lead to stronger perceived rights over land of the people.

This made people more willing to spend on permanent materials for their houses. This is

confirmed by Nyametso (2012) studying poor settlements in Accra, Ghana in terms of

impacts of both legal and de facto customary tenure security. He proved that similar to

12

legal, de facto customary play an important role in encouraging people’s willingness to

make investment on housing while not sufficient enough to be the cause of improvement.

Return to investment - Improved housing as an asset

The second mechanism is that tenure security can mean improved transferability

and values of housing as an asset thus incentivizing people to invest more. Formalization

of land tenure can lead to a better functioning housing market (De Soto, 2000; Durand-

Lasserve and Selod, 2007). With legal recognition, the uncertainty of informal housing

market is reduced making it more appealing to investors and also other consumer (Feder

and Feeny, 1991). This turns housing into a tradable, more valuable asset that they can

further invest on and hope for a future return. Moreover, formalization of land tenure often

comes with greater transfer rights. The additional ability to transfer the land encourages

people to increase their investment. Liu et al. (1998) looked at Chinese communities with

different levels of transfer rights. They concluded that the application of green manure (one

form of investment) increased as the level of transfer rights increase. We can see that the

improvement in exchange value of property (either through better functioning market or

greater capacity of people to join the market with transfer rights) can be the way that land

tenure influences the housing investment.

Not only the transferability of the asset but also the better access to infrastructure,

services especially from the government due to increased tenure security allow increase in

housing investment. People having legal tenure tend to be more qualified for government

assistance. They also have better chance to ask for installation of basic services and

infrastructure (Nakamura 2017, Durand-Lasserve and Selod, 2007). This enhanced access

13

to services and infrastructure has positive impacts on investment in housing (Strassmann.

1984).

Some critique of the improved property rights approach have pointed out negative

effects of the market on life of the people. Although the land and housing markets become

more efficiency and profitable leading to the increase in values of the poverty, the higher

rent and prices can make it harder for the poorest groups to access land and housing in the

areas (Payne, 2004; Gilbert, 2002). As the result, they get pushed out of the community.

Access to credit

Another impact of land tenure security especially through legal security

improvement interventions like formalizing tenure is the increase in access to credit. De

Soto (2000) argued that the legalized rights to the property will help turn it into a live form

of capital. Recorded title allows people to use their house and land as collateral, making it

possible and easier to access credit from the bank. They thus have bigger capital flow to

make investment in either housing and other consumptions and activities (Feder, 1988).

There are, however, mixed results from previous studies exploring the impact of

land tenure security on access to credit. While some found positive relationship between

the two, there are often conditions in order for significant impacts to be obtained. Goyal

and Deininger (2010) studying communities in India have observed the effects of land

registry on borrowing only in urban areas. Feder and Nishio (1998) provided evidence that

several factors, including robust financial market, incentives in land investment, demand

for transactions, and a working system for land registration are required in order to see

some positive impacts of titling on credit access. Feder (1988) highlighted that for areas in

14

which people base more on the informal credit market, which values personal relationship,

the difference in credit level with or without title is not significant. There are also studies

in which no or very little effect on credit access can be found in places where formal credit

markets are not present (Carter and Wiebe, 1994). Carter et al. (1996) found out that the

size of the property is an important condition to evaluate the impact of title on credit access.

They observed that smaller-sized property is already excluded from the credit market thus

no effect can be seen due to the title.

Despite the mixed results, in the debate over tenure security and housing

investment, significantly positive relationships are more commonly found. As a result,

interventions focusing on ensuring tenure security such as providing people with titles or

long-term leasehold are becoming more and more popular to solve the issue of low housing

quality and living condition. The government and policy makers’ interventions seem

crucial in order to improve the housing condition especially in the poor community.

In the three mechanisms explained above, we can see the important role of the

ability to participate and benefit from the market as a reason why greater tenure security

can influence housing investment. However, other factors besides tenure security can drive

this ability and in turns affects the level of housing investment. Proponents of market

approach to land allocation believe that in a competitive market the exchange value of land

or housing that the land owner enjoys can reflect that use values which can be gained by

the highest bidder renting or buying the property. Hence this focus on the exchange value

leads to the higher incentive to invest to maintain or improve the property which can be

exchanged and benefited from (Pivo, 1984).

15

We can see an interaction between the government and the market force here in

terms of housing investment. While studies focus a lot on how tenure security can influence

housing investment through impacting the transferability and the exchange value of the

property, interventions in poor communities directly focusing on the market and how

people participate in it seem to draw less attention. However, in order to better understand

tenure security and how important it is in encouraging housing investment and

improvement, the fact that people do not only enjoy use value but also exchange value

from their property is important to consider.

3.4. Summary of literature review

Previous studies have shown that numerous factors influence the values of land,

whether it is use or transfer values. One of the important factors determining land values

is land tenure. Besides de jure or legal tenure, de facto and perceived tenure security are

also important aspects of tenure security. There are mixed findings about how land tenure

security is related to housing investment and improved housing quality, which is critical,

especially in terms of poverty reduction policies.

16

Chapter 4. RESEARCH DESIGN

In order to explore the relationship between tenure security and housing investment

in Bangladeshi poor communities, a cross-sectional study of the relationship using a multi-

level model of household and community data is utilized. This section will explain the

details about data collection, as well as the statistical model used for analysis.

4.1. Data Collection

Cities in Bangladesh are divided officially into different wards, which are the

smallest administrative urban geographic unit. Each ward is then informally divided into

mahalla, which are the lowest urban geographic units having identifiable boundaries. Poor

settlements are geographically identifiable areas comprised of a group of households with

one or more of the following characteristics: (i) poor quality housing; (ii) extremely high

risk of flooding; (iii)lack of access to potable water and bathing facilities; (iv) lack of

sanitation facilities; (v) insecure land tenure; (vi) high density slums; (vii) inadequate solid

waste management; (viii) lack of electricity; and (ix) lack of access roads and drainage.

This study looks households in the settlements of Chandpur and Narayanganj. Two

different datasets are available from the National Urban Poverty Reduction Program

(NUPRP) in Bangladesh: the household survey and the urban poor settlement assessment.

The urban poor settlement assessment provides the geographical boundaries of poor

settlements which have not been officially defined before and assessment of the

settlements’ conditions such as social infrastructure and social-economic status. This

assessment is conducted in a six-step participatory urban poor settlement mapping process

as shown in Figure 4.1.

17

After necessary materials such as city maps, ward maps, mahalla maps, and score

cards listing indicators of settlements’ socio-economic and physical information have been

prepared in step 1, a consultation workshop is done in the second step to get consent from

the Mayor for the Mapping, and most importantly to identify the community surveyors

involving in the later steps with recommendation from the councilors. Surveyors are local

people from the poor settlements who meet the requirements regarding education level and

experiences in the survey. In each ward, three survey teams of 2 surveyors are selected

from different parts of the ward. In step 3, a ward level participatory workshop is conducted

in order to train the surveyors to conduct the mapping and collect the data for the poor

settlement assessment. In step 4, the community surveyors do transect walks of the

settlements asking people to identify the boundaries of the settlements and score the

Figure 4.1. Steps of Urban Poor Settlement Mapping

(NUPRP 2017)

18

settlement’s indicators. Five to ten percent of the settlements will be cross-checked in step

5 to ensure the quality of the surveying process. All the collected information is then put

into a database and digitized into maps of poor settlements. In step 7, a validation workshop

is held to share the collected information with the mayor, councilors, key slum

development staff, representatives from NGOs, community leaders, mahalla leaders and

federation leaders. Feedback is collected to finalize the maps and data for the assessment.

The final dataset, referred to as a score card, consists of information about

population, number of households in each settlement, age of the settlement, and 14 poverty

indicators. Population indicator calculates the number of people living in each settlement.

Age of settlement refers to the number of years that the settlement has been in existence.

Each of the 14 poverty indicators is ranked on a 1-4 scale, with 1 considered the poorest

condition and 4 considered the best. The dataset covers all settlements (2323) from two

cities Chandpur (439) and Narayanganj (1884.)

The second dataset that I use is the household survey done by NUPRP. This is a

door-to-door survey that includes every household living in poor settlements. This survey

is done in three steps. First of all, member of the community development centers come to

all the settlements and tell people in advance about why and how the surveys will be done.

Then trained surveyors will conduct door-to-door survey using mobile devices, such as

mobile phones or tablets, to input the information from the households. Surveyors and

community leaders will then recheck to make sure all the households have been

interviewed.

19

The resultant dataset from the household survey consists of information about the

household head (age, gender, religion, education, employment etc.), information about the

household in general (number of family members, children in the house etc.), and a multi-

dimensional poverty index (floor material, ownership of the house, toilet facilities etc.) The

dataset includes 60705 households in two cities Chandpur (23062), and Narayanganj

(37643).

Possibly due to the inputting process, some of the observations do not contain

information about the settlement name associated with the household, making it impossible

to match them with settlements and look at the group effect for those observations. Hence,

I drop these observations to have a better estimating model. A total of 2840 observations

(4.7%) out of 60705 were dropped.

4.2. Data Limitation

Both datasets have large observation size. Data collection is also done

systematically and rechecked to ensure accuracy. However, there are still limitations to be

aware of about the data collection in order to have a good analysis. The use of mobile

devices to collect the data might result in some mistakes during the process of asking

questions, answering, receiving information and inputting it into the system. Moreover,

data can be lost while inputting, processing and storing on the server. However, due to the

rechecking process, the chance of this is minor compared to the large dataset.

4.3. Conceptualization

20

The most important concept for this research is “land tenure security.” It is defined

by the NUPRP as the state of having the protection from involuntary and arbitrary eviction.

A sub-concept of land tenure security is “tenure,” which indicates the relationship between

tenant and landlord or property owner. There are two different dimensions of land tenure

security which are de jure and de facto security. De jure security means the legal rights to

the land or security due to the legal title over the land or property. De facto tenure security,

on the other hand, shows the perceptions of people about their control over the land, which

is not due to legal status. Both de jure and de facto security contribute to people’s

perceptions about their land tenure security, which is the estimated risk of being

involuntarily evicted from their own land. For people squatting on their land, factual

situations (de facto) such as the community recognition will be the only sources of tenure

security. For people owning and renting their property, legal (or de jure) rights will add to

their sense of tenure security. Land tenure security will be approached through people’s

ownership of their property, whether they squat, rent or own their living space.

Another major concept is housing investment. It is the level of capital that gets

spent by the household in maintaining and improving their living space and structure. The

most common way to analyze this is through the physical structure of the housing itself.

The picture below (Figure 4.2) shows one of the popular housing typologies that we

observed in Bangladeshi poor communities during our trip there. They are in temporary

condition with bamboo or tin walls and roofs. The more permanent structures were built

out of concrete or bricks, which also cost a lot more.

21

Figure 4.3 below shows the conceptual model looking into the influence of tenure

security on housing investment. Tenure security influences housing investment, as shown

through the physical housing structure. However, other factors both at household level and

settlement level need to be taken into account to accurately estimate the effects of tenure

security, which are the primary interests. More details about how each concept is

constructed and included in the model come in the next section.

Figure 4.2. Tin Houses in Bangladeshi Poor settlements

22

Figure 4.3. Conceptual model of housing investment

4.4. Variable Descriptions

Dependent variable: Cement floors

To measure the level of investment in housing improvement, I use information

about floor materials from the household survey. This information is inputted by the

interviewer observing the main material for the dwelling floor in each household. Possible

answers are: Cement, Earth/sand, Dung, Wood planks, Palm bamboo, and Others. From a

general sense as well as our visit to several settlements in both cities of interest, I assume

houses with cement floors are more stable and of better quality and condition. Moreover,

considering the regular flooding and water logging issues in Bangladeshi cities, it is

Housing Investment

- Floor Material

Tenure security

- Ownership

Settlement Controls

- Distance to city center, railroad, river

- Road condition

Household Controls

- Use of TV

- Number of dependents

- Female head

23

obvious that cement is more of a resistant material than other commonly used floorings

such as wood, bamboo, or earth/sand. As a result, in order to make it easier for building a

model of analysis, I turn the information into a binary variable named cement with two

possible values of 1 (meaning cement is used for flooring) and 0 (meaning cement is not

used for flooring). For an amended model, I also construct an ordinal variable material

with the value of 3 for cement floors, 2 for wood and bamboo floor, and 1 for the chance

that there can be meaningful difference in investment between wood/bamboo and other

types.

Independent variables:

i. Tenure Security

To measure the level of de jure tenure security, I use information from the

household survey. The survey asks whether the household owns or rents the house they are

living in and whether their property is on government land. Possible answers are: own,

rent on non-government land, rent on government land and squat. Four binary variables

are created to indicate this information. Variable squat is used as reference variable.

ii. Household characteristics control variables

Since household characteristics can influence both the level of investment in

housing and their tenure security, I include some control variables to better indicate the

relationship between the variables of interest.

The first variable is the household head’s employment. From the household survey,

the employment of the household head is asked. A dummy variable hh_employed takes

the value of 1 if the household head has some kind of employment and 0 if the head

24

identifies him/herself as unemployed. Households with employed heads might have better

sources of income leading to their greater ability and willingness to spend more on their

house.

A dummy variable, female, is also included to take into account the household head

gender. It takes the value of 1 if the head is female and 0 if the head is male.

Income level is an important control factor to consider since it can influence the

capacity of the household to invest in their housing. Although I do not have a direct

measure of how much each household earns to control for income, I believe expenditure,

particularly basic expenditures like feeding children and caring for the elderly, can be a

good proxy for income level. The more people have to spend on other needs of the family

like food, transportation or education, the less spare money they have to invest in housing.

I add a variable of number of dependents counting children under 5 and disabled family

members as a proxy for income.

Not only reliable income but also wealth level is an important control variable

influencing investment. Different from income, wealth of a family can be inherited from

parents or gained from other sources. Assets that the households have can reflect their

wealth. I create a dummy variable showing whether the household has television, which

can be considered a luxury asset, in order to control for household wealth.

iii. Settlement control variables

I also add three other variables to control for the location of the settlement that the

household is in. People living closer to city center, river or railroad might have better access

to more stable materials like cement as the transportation cost can be higher for those living

25

very far away. Two dummy variables were added to indicate whether the settlement is

within 1 km from the railroad or the river. Continuous variable (dist_cen) is used to

calculate the distance (in km) from the settlement to the district center taken from the

zoning map.

Access to infrastructure is also an important factor influencing both the tenure

security of people living in the area and the likelihood of investing in their structure. Areas

with better infrastructure tend to have a lower risk of eviction. A more secure property also

encourages people to invest more in it since they are not afraid of the return to their

investment. Hence, I add a ranked variable (road) to account for the condition of access

roads in the settlement. This information is taken from the settlement survey.

Table 4.1. List of variables

Variable Label Definition Source

Dependent

Variable

Floor material

Whether the

household use

cement floors

Cement

1 – The

household use

cement floors

0 – The

household not

use cement

floors

Household

Survey

Independent

Variable

Whether the

household squat

their house or

not

Squat

1 – The

household

squat their land

0 – The

household not

squat their land

Household

Survey

1 – The

household rent

their land from

government

26

Tenure security

Whether the

household rent

their house on

government

land

Rent_gov

0 – The

household not

rent their land

from

government

Household

Survey

Whether the

household rent

their house on

non-

government

land

Rent_non_gov

1 – The

household rent

their land from

non-

government

entity

0 – The

household not

rent their land

from non-

government

entity

Household

Survey

Whether the

household own

their house

Own 1 – The

household own

their land

0 – The

household not

own their land

Household

survey

Settlement

control

variables

Condition of

access road

infrastructure in

the settlement

Road 1–no proper

access road

2-Earth/Gravel

with poor

maintenance

3– Paved

without proper

side drains and

poor

maintenance

4- Paved with

proper side

drains and well

maintenance

Settlement

Survey

Distance from

the settlement

to the center of

the city

Dist_cen

Distance to the

city center in

km

Settlement

Survey

Whether the

settlement is

within 1km

1 – The

settlement is

within 1km

27

from the

railroad

Rr_1km

from the

railroad

0 – The

settlement is

more than 1km

from the

railroad

Settlement

Survey

Whether the

settlement is

within 1km

from the river

River_1km

1 – The

settlement is

within 1km

from the river

0 – The

settlement is

more than 1km

from the river

Settlement

Survey

Household control

variables

Whether the

household has

television

TV

1 – The

household has

TV

0 – The

household does

not have TV

Household

Survey

Number of

dependent

Dependent

Number of

children under

5 and disabled

members

Household

Survey

Whether the

head of the

household is

female

Female

1 – The

household head

is female

0 – The

household head

is male

Household

Survey

4.5. Statistical Procedure

I use statistical software Stata 13.0 to process and run analysis models. Quantitative

analysis is taken as the main approach to look into impacts of tenure security on investment

in housing in settlements in the two Bangladeshi cities using the dataset from NUPRP

28

collected in 2016. Since the dependent variable is a binary variable, logistic regression is

conducted to identify factors influencing the likelihood of having cement floors.

The data is taken from two nested datasets with household level and settlement

level data. This leads to the concern about the similarity of households within the same

settlement. The effects of being in a cluster can lead to the misestimation of standard errors

(Hox, 1998). This concern about context effects have driven the growing trend of

multilevel estimation strategies. There are many different statistical tools that have been

developed and used by various fields to deal with this issue of clustered/nested data such

as mixed-effects multilevel or hierarchical model from sociology, cluster-robust standard

errors used by economists, or two-step approach popular in political science (Heisig et al.,

2015). Although not all the nested data automatically asks for multilevel modelling,

especially in cases where the primary interest is in level 1 effects and when observations

are relatively independent (intra-class correlation is small), accounting for clustering

effects is always important in order to produce “honest models” (Raudenbush & Bryk,

2012.)

This study utilizes a popular multilevel estimation model, which is Cluster-robust

standard errors for logit model. With a large enough number of clusters, or in this case

settlements (over 50), this model gives quite reliable estimates of the relationship between

variables in a hierarchical dataset. Moreover, unlike other models such as mixed effect and

two-step approaches, the processing, as well as interpretation of results from this model is

much simpler allowing it to be more accessible to people of different fields.

Cluster-robust standard errors logistic model

29

The difference between the cluster-robust standard error method with the

conventional logit model relies on the corrected standard errors estimates to account for

the cluster setting (Heisig et al. 2015.) The resultant standard errors will be substantially

larger than the ones from traditional logistic models. The model will look similar to the

traditional logistic regression models. Model (1) shows the model used for the study.

Y = B0 + B1Z + B2Q + B3T + u (1)

Y = Odd ratio showing likelihood of the household having cement floors

B0, B1, B2 = coefficient variables

Z = Tenure security variables

Q = Settlement control variables

T = Household control variables

u = cluster-robust standard error

Interpretation of the logistic regression model is different from the Ordinary Least

Square (OLS) model. The easiest way to read the coefficients is by the “odds ratio.” Odds

ratio is obtained by taking the exponent constant (~ 2.72) and raising it to the power of the

coefficient Bi. An odds ratio of more than 1 implies a positive relationship. For example,

an odds ratio of 1.20 can be read as 1 unit increased in the independent variable results in

a 1.20 time increase in likelihood that dependent variable equals 1. An odds ratio lower

than 1 display a negative relationship. For example, an odds ratio of 0.70 means that 1 unit

increased in the independent variable results in 30% decrease in likelihood that the

dependent variable equals to 1.

30

Although quite flexible and easy to use, cluster-robust standard errors model also

has some potential drawbacks. The most important weakness of this model is that a large

number of clusters or groups are required for accurate estimations. The suggested size is

50 or more clusters (Kezdi 2004; and Heisig 2017.) When the number of clusters falls

below the suggested level, other methods or corrections are needed. Since the cluster

number is not a problem for these datasets, cluster-robust standard error is a quite useful

and appropriate method to use.

31

Chapter 5. RESULTS

5.1. Descriptive statistics

This subsection gives a brief overview of all observed households in terms of their

housing material (floor material) and other factors including household characteristics and

the settlements that they live in. Of all 56,334 households in this analysis, 32686 (or 58%)

have cement floors. Table 5.1 below presents a brief summary of all the variables.

Table 5.1. Descriptive statistics

Variable Observations Mean

Standard

deviation min max

Female =1 17080

Cement = 1 32696

Squat=1 8495

rent_gov=1 67

rent_nongov =1 24911

own=1 22861

rail_1km=1 29162

river_1km=1 41503

tv=1 32534

road 56399 1.86 0.946 1 4

dependent 56399 1.08 0.703 0 10

dis_cen 56399 2250.76 2531.62 0 10917.8

32

Table 5.2 shows numbers of household with and without cement floors in terms of

ownership types over the land they are living on. We can see that higher percentages of

households owning and especially renting their house not from the government have

cement floors. In contrast, more than 5,000 out of 8,495 households squatting on their land

do not use cement as their floor material. Among group renting from the government, we

also see a greater portion using floor materials other than cement. This suggests that having

more legal rights to the property has some positive effects on housing structure investment.

Table 5.2. Floor material and Ownership type

Table 5.3 below shows the road conditions that poor households in the two cities

experience. Over 80% of them either have no proper access road or have only earth/gravel

with poor maintenance. From Table 5.4, we can see some relationship between the road

conditions and the use of cement as floor materials. Households with the worst access road

conditions are less likely to have cement floors. Within the group of households having

better road conditions, a higher percentage of them have cement floors. However, the

increase in percentage is a bit small. Moreover, the group with “Paved with proper side

.

Pearson chi2(3) = 2.6e+03 Pr = 0.000

100.00 100.00 100.00 100.00 100.00

Total 8,495 67 24,911 22,861 56,334

39.72 46.27 68.86 53.05 58.02

Yes 3,374 31 17,154 12,127 32,686

60.28 53.73 31.14 46.95 41.98

No 5,121 36 7,757 10,734 23,648

cement squat rent-gove rent-nong own Total

ownership

33

drains and well maintenance” roads actually sees a lower percentage of cement-floor

houses than the one with “Paved without proper side drains and poor maintenance” road

condition. Considering this along with other factors might help clarify whether there is any

impact of infrastructure (roads in this case) on the choice of housing floor, as well as how

strong that impact is.

Table 5.3. Road condition of in poor households

Table 5.4. Road condition and housing material

A large portion of the households (about 74%) live in settlements which are within

1km of the river. There is no strong relationship between proximity to the river and the

likelihood of using more stable housing materials (cement). Table 5.5 shows that the

Total 56,398 100.00

Paved with proper side drains and well 5,585 9.90 100.00

Paved without proper side drains and po 4,973 8.82 90.10

Earth/Gravel with poor maintenance 21,559 38.23 81.28

No proper access road 24,281 43.05 43.05

road Freq. Percent Cum.

Pearson chi2(3) = 1.9e+03 Pr = 0.000

42.03 57.97 100.00

Total 23,703 32,695 56,398

33.79 66.21 100.00

Paved with proper sid 1,887 3,698 5,585

30.85 69.15 100.00

Paved without proper 1,534 3,439 4,973

34.94 65.06 100.00

Earth/Gravel with poo 7,533 14,026 21,559

52.51 47.49 100.00

No proper access road 12,749 11,532 24,281

road No Yes Total

cement

34

percentage of households with cement floors within 1km from the river is a bit lower (2%

lower) than that of the group living further away.

Table 5.5. Proximity to river and housing material

Proximity to the railroad seems to have some relationship to the likelihood of

households using cement floors. From Table 5.6 below we can see that for the group living

within 1 km of the railroad, a much higher percentage have cement floors (63%) compared

to households living further away (52%).

Table 5.6. Cement and proximity to the railroad

Pearson chi2(1) = 23.1417 Pr = 0.000

100.00 100.00 100.00

Total 14,896 41,503 56,399

56.30 58.57 57.97

Yes 8,387 24,309 32,696

43.70 41.43 42.03

No 6,509 17,194 23,703

cement No Yes Total

Within 1km from river

Pearson chi2(1) = 636.6435 Pr = 0.000

100.00 100.00 100.00

Total 27,237 29,162 56,399

52.55 63.04 57.97

Yes 14,312 18,384 32,696

47.45 36.96 42.03

No 12,925 10,778 23,703

cement No Yes Total

railroad

Within 1km from the

35

Besides the settlement characteristics, household features like number of

dependents, the use of television, and the gender of the head can also impact the level of

investment in housing. Of all the households, about 58% have television. For those with

television, a quite high percentage own cement floors (about 70%). We can also see from

Table 5.7 that among households without a TV, the percentage using cement floors drops

to about 41%. This confirms our asumption that wealthier families (having TV) have more

resources to invest in their housing as well.

Table 5.7. Cement and the use of TV

Most households (about 97%) in the study area have three or fewer dependents,

calculated as number of children under 5 and disabled members. The Pearson’s correlation

of 0.164 suggests a weak and positive relationship between the number of dependents and

the use of cement floors. This conflicts with the assumption that more dependents would

require greater expenditure on basic needs taking away resources for housing

improvement. This weak positive relationship might reflect the fact that dependents work

and support their family instead of being only a burden.

Pearson chi2(1) = 4.7e+03 Pr = 0.000

100.00 100.00 100.00

Total 23,865 32,534 56,399

41.38 70.15 57.97

Yes 9,875 22,821 32,696

58.62 29.85 42.03

No 13,990 9,713 23,703

cement No Yes Total

tv

36

From the descriptive statistics, we can see some correlations between the use of

cement and factors like ownership over the property, use of TV, proximity to river or

railroad, as well as road conditions. However, digging further into the data with the

proposed multilevel estimation is important for more reliable conclusions. The results of

this model will be presented in the next subsection.

5.2. Cluster-robust standard errors logistic regression model

Table 5.7 below shows the results from running a logistic regression corrected with

cluster robust standard errors. The coefficients have been calculated by the tool and

displayed in odds ratio for easier interpretation. From the table, we can see that renting

from government and own variable is significant at 95% significance level. People who

own are about 1.4 times more likely to have cement floors than those who squat.

Surprisingly, people who rent their property not from the government are about 2.8 times

more likely to have cement floors than those who squat, which is an even higher likelihood

than people owning their place. A similar model ran with owner occupants as the reference

group shows that renters of non-government properties are about twice as likely than owner

occupants to have cement floors. Those who rent from the government don’t see any

significant difference from squatters in terms of the floor materials they use. This result

suggests that greater legal rights from owning or renting houses on the private market do

have positive impact on the likelihood of having a more permanent housing structure. Since

the number of households renting from the government is very small (67) compared to

other types of ownership, the evidence is not reliable to draw any conclusion about this

group.

37

Among the control variables, both the household characteristics (number of

independent and possession of TV, the type of land they live on) and neighborhood

characteristics (proximity to railroad, distance to city center) are significant variables.

Households with more dependents are more likely (1.35 times) to have cement floors. This

is contradictory to the assumption that more dependents mean higher basic expenditures

leading to less available resources to spend on housing. This might be due to the fact that

dependents like children and disabled people are actually the human capital working to

support their family more than an additional burden on other members. Compared to the

number of dependents, TV, the indicator of wealth has much stronger influence on the use

of cement floors. Those with a TV are 3.27 times more likely to also invest in their cement

floors.

Access to road infrastructure and railroad are very significant factors determining

the likelihood of the household having cement floors. An increase in quality of road

condition results in an increase of 1.26 times in the likelihood of the household having

cement floors. Households living in the settlement within 1km of the railroad experience

an increased likelihood (1.37 times) to have cement floors. These findings align with the

hypothesis that better neighborhood conditions like access to infrastructure and amenities

can encourage people to invest more in their own properties.

38

Table 5.8. Logistic regression with cluster-robust standards error results

For the purpose of comparison, I also run the data again using mixed-effect model,

which I don’t go into great details to account for cluster effects. Results from this method

also show quite similar patterns as the cluster-robust standard error models. However, in

this model, the effect of renting from government land also becomes significant. Moreover,

the proximity to the river has significant negative relationship with having cement floors.

Table 5.9 shows the results from the mixed-effect model.

_cons .193515 .0492912 -6.45 0.000 .1174629 .3188075

rail_1km 1.366076 .1744294 2.44 0.015 1.063624 1.754534

river_1km .9169314 .123137 -0.65 0.518 .7047366 1.193018

dis_cen .9999606 .0000184 -2.14 0.032 .9999245 .9999967

road 1.255677 .0612321 4.67 0.000 1.141221 1.381612

dependent 1.346515 .090529 4.43 0.000 1.180275 1.53617

tv 3.273312 .2236047 17.36 0.000 2.863127 3.742262

female .9576896 .0369476 -1.12 0.262 .8879439 1.032914

own 1.413721 .2397715 2.04 0.041 1.013905 1.971197

rent_nongov 2.765697 .4786396 5.88 0.000 1.970122 3.882541

rent_gov 1.459039 .5284175 1.04 0.297 .7174454 2.967188

cement Odds Ratio Std. Err. z P>|z| [95% Conf. Interval]

Robust

(Std. Err. adjusted for 1792 clusters in settlement)

Log pseudolikelihood = -33976.013 Pseudo R2 = 0.1146

Prob > chi2 = 0.0000

Wald chi2(10) = 685.69

Logistic regression Number of obs = 56398

39

Table 5.9. Mixed-effects logistic regression result

LR test vs. logistic regression: chibar2(01) = 13776.15 Prob>=chibar2 = 0.0000

sd(_cons) 1.775377 .0446469 1.689992 1.865076

settlement: Identity

Random-effects Parameters Estimate Std. Err. [95% Conf. Interval]

_cons .2706183 .0396221 -8.93 0.000 .2031093 .3605657

rail_1km 2.032472 .2083423 6.92 0.000 1.662535 2.484727

river_1km .6789124 .0798245 -3.29 0.001 .5391779 .8548609

dis_cen 1.000012 .0000182 0.64 0.520 .999976 1.000047

road 1.324612 .0614352 6.06 0.000 1.209512 1.450666

dependent 1.096145 .0218257 4.61 0.000 1.054192 1.139769

tv 3.341757 .0827371 48.73 0.000 3.183467 3.507918

female .9249137 .0233128 -3.10 0.002 .8803317 .9717535

own 1.437007 .0707999 7.36 0.000 1.304731 1.582693

rent_nongov 1.899703 .0897874 13.58 0.000 1.731628 2.084092

rent_gov 2.1101 .6516096 2.42 0.016 1.151987 3.865079

cement Odds Ratio Std. Err. z P>|z| [95% Conf. Interval]

Log likelihood = -27087.938 Prob > chi2 = 0.0000

Integration points = 7 Wald chi2(10) = 2667.83

max = 1718

avg = 31.5

Obs per group: min = 1

Group variable: settlement Number of groups = 1792

Mixed-effects logistic regression Number of obs = 56398

40

Chapter 6. DISCUSSION

From the statistical results, we can see that greater tenure security can mean more

likelihood of having cement floors in poor households in the two Bangladeshi cities. Those

renting or owning their property are more likely to invest in their house than squatters. This

finding again proves the possible positive impacts of interventions targeting tenure security

in order to encourage housing investment leading to better housing and living conditions.

However, unlike what I assume, people who own the house are actually less likely

to have cement floors than renters of non-government properties. This leads back to the

discussion about use value vs. exchange values of the property. For those owning the house,

they mostly enjoy the use values of the property. Poor people might have other priority for

expenditure such as food or schooling of their kids than improving the living space. The

rental units can be seen as a form of business investment. As a result, increase in exchange

values of the properties can encourage the actual owner of the units to spend more on the

structure, especially in the case of non-government land. This case of household in

Bangladeshi poor communities has shown that the ability to participate in the housing

market and benefit from the property can significantly influence the housing investment.

Although some stable rights to the property either through renting or owning do

have positive impacts on the likelihood of having more permanent housing structure, there

are also other very important factors influencing the investment in housing. One interesting

finding is the importance of road conditions and proximity to the railroad. This finding can

support the argument that the ability to benefit from the exchanging (or renting in this case)

of the house encourages the owners of the unit to invest more on the structure.

41

Neighborhoods with better conditions such as access to railroad or more developed road

system tend to have higher property value. More people are willing to buy or rent houses

in these areas due to the accessibility and convenience driving up the exchange values.

Property owners then have more incentive to keep the structure maintained and even

improved with more permanent materials like cement.

Results from this research confirm positive impact of increased legal rights on

housing condition, shown through physical housing structure. However, the findings also

suggest even a stronger impact of returns to investment through the increase in exchange

values on housing condition. Rental units or in properties with good access to infrastructure

like road or railroad are good opportunities, which can promise greater return to

investment. Besides the impacts from improved tenure security, the direct effects due

benefiting from the housing market on investment in housing is considerable. The next

chapter will go deeper in the implications of these findings for policy makers as well as

future research.

42

Chapter 7. IMPLICATION

For policy makers:

The findings suggest that allowing greater sense of tenure security through having

rental contract or title might encourage people to improve their physical living structure.

However, the finding of greater likelihood of renters of nongovernment units than owner

occupants to have cement floors suggests that housing can also be seen as a form of

business opportunity beside a living space. This could be meaningful for policies making

in terms of economic development. For the population groups such as elderly or disabled

people, who struggle to benefit from many other economic development efforts such as

creating jobs or training, having chance to benefit from their own housing asset might be

very important. The findings of importance of access to railroad and road infrastructure

also contribute to this idea of looking at housing as not only a living space but also a

business opportunity. Lifting neighborhood characteristics such as access to infrastructure

and amenities can contribute to greater investment in housing since people can see a greater

chance of good return to their investment through either selling or renting out their

property.

However, whether the focus is on improving people’s living quality in terms of

housing quality and quantity or on helping people to have greater business opportunities

will influence the way policies are directed to. Future interventions should be aware of and

balance both of these two aspects of housing in order to be more effective in terms of

improving people’s life especially in poor urban communities.

43

Not only the government but also other agency working on improving housing and

living conditions of the poor in Bangladesh can utilize the results from this research. The

NUPRP, who provide datasets enabling the conduct of this study, has been working on and

advocating for comprehensive solutions for urban issues in Bangladeshi poor settlements.

From the findings we can see that improving neighborhood characteristics such as road,

drainage systems or access to other amenities can have impacts on housing conditions.

Hence this can be considered for effective solutions.

For future research:

While the current literature about housing in poor communities heavily focuses on

providing better quality and quantity of housing for people, this research suggests that

market forces also influence people’s decision in terms of their housing. It is important for

future research, especially on poor communities to also take into account this aspect of

housing in order to better understand the issues.

The suggested model seems to be meaningful in providing more insights about the

pattern of housing investment in poor communities in the two cities of Narayanganj and

Chandpur. It might be useful for future research to consider this to explore the same issue

in the context of other cities in Bangladesh as well as other countries especially in the

developing world. However, additional information about differences in cultures, size, the

market and political systems to name just a few are very important if this model and results

are considered in other contexts.

44

REFERENCES

1. Ahmed, K. I. (2007). Urban poor housing in Bangladesh and potential role of ACHR.

Bangkok: Asian Coalition for Housing Rights (ACHR)

2. Besley, T. (1995). Property rights and investment incentives: Theory and evidence

from Ghana. Journal of Political Economy, 103(5), 903-937.

3. Brasselle, A., Gaspart, F., & Platteau, J. (2002). Land tenure security and investment

incentives: Puzzling evidence from Burkina Faso. Journal of Development Economics,

67(2), 373-418.

4. Carter, M. R., Fletschner, D., & Olinto, P. (1996). Does land titling activate a

productivity-promoting land market? Econometric Evidence from Rural Paraguay',

University of Wisconsin-Madison,

5. Carter, M. R., & Wiebe, K. D. (1990). Access to capital and its impact on agrarian

structure and productivity in Kenya. American Journal of Agricultural Economics,

72(5), 1146-1150.

6. Clark, D. (1982). Urban geography: An introductory guide Taylor & Francis.

7. Coulton, C. J., & Pandey, S. (1992). Geographic concentration of poverty and risk to

children in urban neighborhoods. American Behavioral Scientist, 35(3), 238-257.

8. Dales, J. H. (1968). Land, water, and ownership. The Canadian Journal of

Economics/Revue Canadienne d'Economique, 1(4), 791-804.

9. De Soto, H. (2000). The mystery of capital: Why capitalism triumphs in the West and

fails everywhere else. New York: Basic Books.

10. Deininger, K., & Feder, G. (2001). Land institutions and land markets. Handbook of

Agricultural Economics, 1, 287-331.

11. Demsetz, H. (1974). Toward a theory of property rights. Classic papers in natural

resource economics (pp. 163-177) Springer.

12. Durand-Lasserve, A., & Selod, H. (2009). The formalization of urban land tenure in

developing countries. Urban Land Markets: Improving Land Management for

Successful Urbanization, 101-132.

13. FAO. (2002). [Land Tenure and Rural Development] FAO Land Tenure Studies, 3

14. Feder, G. (1988). Land policies and farm productivity in Thailand Johns Hopkins

University Press.

15. Feder, G., & Feeny, D. (1991). Land tenure and property rights: Theory and

implications for development policy. The World Bank Economic Review, 5(1), 135-

153.

16. Feder, G., & Nishio, A. (1998). The benefits of land registration and titling: Economic

and social perspectives. Land use Policy, 15(1), 25-43.

17. Field, E. (2005). Property rights and investment in urban slums. Journal of the

European Economic Association, 3(2‐3), 279-290.

18. Galiani, S., & Schargrodsky, E. (2010). Property rights for the poor: Effects of land

titling. Journal of Public Economics, 94(9), 700-729.

19. Gilbert, A. (2002). On the mystery of capital and the myths of Hernando De Soto: What

difference does legal title make? International Development Planning Review, 24(1),

1-19.

45

20. Goyal, A., & Deininger, K. (2010). Going Digital: Credit Effects of Land Registry

Computerization In India. Policy Research Working Papers.

21. Harvey, D. (2010). Social justice and the city University of Georgia Press.

22. Heisig, J. P., Schaeffer, M., & Giesecke, J. (2015). The Costs of Simplicity: Why

Multilevel Models May Benefit from Accounting for Cross-Cluster Differences in the

Effects of Controls. American Sociological Review, 82(4), 796-827.

23. Hox, J. (1998). Multilevel Modeling: When and Why. Classification, Data Analysis,

and Data Highways Studies in Classification, Data Analysis, and Knowledge

Organization, 147-154.

24. Kezdi, G. (2004). Robust Standard Error Estimation in Fixed-Effects Panel Models.

SSRN Electronic Journal.

25. Liu, S., Carter, M. R., & Yao, Y. (1998). Dimensions and diversity of property rights

in rural china: Dilemmas on the road to further reform. World Development, 26(10),

1789-1806.

26. Nakamura, S. (2014). Impact of slum formalization on self-help housing construction:

A case of slum notification in India. Urban Studies, 51(16), 3420-3444.

27. Nakamura, S. (2016). Revealing invisible rules in slums: The nexus between perceived

tenure security and housing investment. Habitat International, 53, 151-162.

28. Nakamura, S. (2017). Does slum formalization without title provision stimulate

housing improvement? A case of slum declaration in Pune, India. Urban Studies, 54(7),

1715-1735.

29. Nyametso, J. K. (2012). The link between land tenure security, access to housing, and

improved living and environmental conditions: A study of three low-income

settlements in Accra, Ghana. Norsk Geografisk Tidsskrift-Norwegian Journal of

Geography, 66(2), 84-98.

30. Payne, G. (2004). Land tenure and property rights: An introduction. Habitat

International, 28(2), 167-179.

31. Pivo, G. E. (1984). Use value, exchange value, and the need for public land-use

planning. Berkeley Planning Journal, 1(1)

32. Place, F., Roth, M., & Hazell, P. (1994). Land tenure security and agricultural

performance in Africa: Overview of research methodology. Searching for Land Tenure

Security in Africa, 15-39.

33. Raudenbush, S. W., & Bryk, A. S. (2012). Hierarchical linear model applications and

data analysis methods. Thousand Oaks: Sage.

34. Spencer, J. H. (2004). People, places, and policy: A politically relevant framework for

efforts to reduce concentrated poverty. Policy Studies Journal, 32(4), 545-568.

35. Strassmann, W. P. (1984). The timing of urban infrastructure and housing

improvements by owner occupants. World Development, 12(7), 743-753.

36. The World Factbook: BANGLADESH. (2018, April 09). Retrieved from

https://www.cia.gov/library/publications/resources/the-world-factbook/geos/bg.html

37. Van Gelder, J. (2010). What tenure security? the case for a tripartite view. Land use

Policy, 27(2), 449-456.

46

38. Van Gelder, J., & Luciano, E. C. (2015). Tenure security as a predictor of housing

investment in low-income settlements: Testing a tripartite model. Environment and

Planning A, 47(2), 485-500.

39. Varley, A. (2013). Post-colonializing informality? Environment and Planning D:

Society and Space, 31(1), 4-22.

40. Walker, R. (1981). Theory of suburbanization: Capitalism and the construction of the

urban space in the united states Methuen.

47

APPENDIX

Logistic regression with owner occupants group as reference

_cons .2725952 .0386219 -9.17 0.000 .2064991 .3598473

rail_1km 1.363209 .1740167 2.43 0.015 1.061462 1.750734

river_1km .9153815 .1230328 -0.66 0.511 .703389 1.191266

dependent 1.346735 .0904665 4.43 0.000 1.1806 1.536248

dis_cen .9999606 .0000184 -2.14 0.032 .9999246 .9999966

road 1.256122 .0612504 4.68 0.000 1.141631 1.382094

tv 3.27865 .2239816 17.38 0.000 2.867775 3.748392

female .9579083 .0369644 -1.11 0.265 .8881313 1.033167

squat .7158663 .1218763 -1.96 0.050 .5127607 .9994225

rent_nongov 1.9649 .1316184 10.08 0.000 1.72315 2.240568

rent_gov 1.136992 .3707693 0.39 0.694 .6000446 2.154425

cement Odds Ratio Std. Err. z P>|z| [95% Conf. Interval]

Robust

(Std. Err. adjusted for 1792 clusters in settlement)

Log pseudolikelihood = -33981.033 Pseudo R2 = 0.1144

Prob > chi2 = 0.0000

Wald chi2(10) = 688.93

Logistic regression Number of obs = 56398